Re: [PATCH net-next v3 4/8] rust: time: Implement addition of Ktime and Delta

[Boqun Feng]

On Thu, Oct 17, 2024 at 08:58:48PM +0200, Miguel Ojeda wrote:
> On Thu, Oct 17, 2024 at 7:03 PM Boqun Feng <boqun.feng@xxxxxxxxx> wrote:
> >
> > but one thing I'm not sure is since it looks like saturating to
> > KTIME_SEC_MAX is the current C choice, if we want to do the same, should
> > we use the name `add_safe()` instead of `saturating_add()`? FWIW, it
> > seems harmless to saturate at KTIME_MAX to me. So personally, I like
> 
> Wait -- `ktime_add_safe()` calls `ktime_set(KTIME_SEC_MAX, 0)` which
> goes into the conditional that returns `KTIME_MAX`, not `KTIME_SEC_MAX
> * NSEC_PER_SEC` (which is what I guess you were saying).
> 

Yeah.. this is very interesting ;-) I missed that.

> So I am confused -- it doesn't saturate to `KTIME_SEC_MAX` (scaled)
> anyway. Which is confusing in itself.
> 

Then I think it suffices to say ktime_add_safe() is just a
saturating_add() for i64? ;-)

> In fact, it means that `ktime_add_safe()` allows you to get any value
> whatsoever as long as you don't overflow, but `ktime_set` does not
> allow you to -- unless you use enough nanoseconds to get you there
> (i.e. over a second in nanoseconds).
> 

That seems to the be case.

Regards,
Boqun

> Cheers,
> Miguel
>